Re: [ql-users] new hard disk
Morning Ade, [EMAIL PROTECTED] wrote: I must admit, I was assuming Sinclair had used 1024 byte blocks on his microdrives - I may need to be corrected on that. I suspect that 1024 is correct. Although, the freespace/total space numbers (on a DIR or STAT) was reported in sectors with a sector being 512 bytes if I remember correctly. I think each file had a 64 byte overhead on the first sector for the file header which was also the directory entry. However, having said that, I have never been able to read the 64 byte file header, or seen it, directly from the file - even though I've seen it written down that it can be done. I'm pretty sure I even used a disc sector editor program to check it out - still no joy. I remain to be convinced of the actual existense of this phantom 64 byte header per file actually 'in' the file. Stephen Usher's description of the perils of formatting is amongst the best I've ever seen. It's true that you can lose staggering amounts of disk space to a bad file format... Yes. I remember the old days when as you added a bigger disc to DOS/Windows, you didn't get as much extra space as you thought. The bigger the disc, the bigger the cluster size so the bigger your small files actually were in reality. This was due to FAT16 (the forerunner to FAT32) only having 16 bit numbers - so if you got too many MB on the new drive, it 'adjusted' the cluster size to allow the whole disc (subject to some other limit) into a 16 bit number. Very helpful indeed - not! However, I think anything in the 3 to 4 million microdrive equivalents will probably last most of us for a while yet (unlike a 400GB PC disk, which at current rates will be obsolete in 18 minutes and 23 seconds). Hmmm. I'm just wondering how long it would take to feed the afore mentions 3.x million cartridges into Dilwyn's Super Disc Indexer/Labeller program to catalogue the contents of them all. Let's see : * assume 30 seconds per scan (that's optimistic!) * assume 3.5 m illion cartridges. So, that's 1.75 million seconds, not including run up/run down and swapping over time. That works out at 29,166 minutes and 40 seconds. That is 486 hours, 6 minutes and 40 seconds or 20 days 6 hours 6 minutes and 40 seconds of continual time. That's a hell of a lot of cartridge labels too :o) Cheers, Norman. ___ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
Re: [ql-users] new hard disk
[EMAIL PROTECTED] wrote: I must try and write a program to see just how many directories there are. Assuming Linux (because you mentioned it) how about : cd \ ls -Rl | grep ^d | wc -l I got 2,889 on a test system I have here at work - and that's not from the root of the drive. Cheers, Norman. ___ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
Re: [ql-users] new hard disk
[EMAIL PROTECTED] wrote: [EMAIL PROTECTED] wrote: I must try and write a program to see just how many directories there are. Assuming Linux (because you mentioned it) how about : cd \ ls -Rl | grep ^d | wc -l I got 2,889 on a test system I have here at work - and that's not from the root of the drive. You'd be better off with cd / find ./ -type d -print | wc -l :-) ___ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
Re: [ql-users] new hard disk
[EMAIL PROTECTED] schrieb: Norman, investigation ist simple: * ls -lR | grep ^d | wc -l 11.735 seconds of real time to find 2,889 directories This will put a directory of _all_ files on your harddisk into the pipe, grep will search through it and throw away most of it. * find ./ -type d -print | wc -l 1.133 seconds to find 3,451 directories. This will only put directory names into the pipe. Command I has produced a whole lot more data that needed to be handled (and most of it thrown away afterwards), whereas command II only produced the data you wanted. But all that is probably way off-topic. Tobias So, not only is find much faster, it finds more directories too. Hmmm. Further investigation required methinks. Cheers, Norman. ___ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm ___ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
Re: [ql-users] new hard disk
David Tubbs wrote: At 15:10 30/04/2007, you wrote: I must admit, I was assuming Sinclair had used 1024 byte blocks on his microdrives - I may need to be corrected on that. 512 byte seectors. one map sector, one byte for each potential sector, I had a few mdvs of 250 sectors. didn't it have 1 word, 2 bytes per sector: the file number in one byte ($f8 = sector map, $fd = free, $ff = dead) plus the block number within the file in the othe byte? Note tpp, every file fragmented of necessity by the interleave factor allowing the QL to digest the data from one sector before reading the next, some 11 or 13 further on. If the file were contiguous it would require a full revolution between sector reads. That begs the question of what one means by fragmented? If the sector allocation is such that some are deliberately skipped (interleave) then surely a fragmented file would be one that doesn't use the preferred sector(s), which for DOS users (with hard disks) would be the next contiguous sector (apparently). I had some of the Psion package which were Turbo Load, laid out for optimum pickup speed, they had to be copied by special procedure equivalent to a DOS DISCOPY. I never did that, but I had heard of it being done - the file laid out so that when the QL had digested the current sector, the next required one would be passing the read head...did it take into account scatter[1] loading? [1] As each sector has a file number and block number, it's position in the file is instantly recognised when read and if a later block happens past the read head before an earlier one, it is loaded first, into the correct memory. ___ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm